Population Pharmacokinetic Analysis Data (ADPPK) Programming in {admiral} and the Pharmaverse

PHUSE US Connect 2024

Jeff Dickinson, Navitas Data Sciences

Abstract

Population Pharmacokinetic modeling is an important tool for drug development. The CDISC ADaM Population PK Implementation Guide is forthcoming. Population PK models generally make use of nonlinear mixed effects models that require numeric variables. The data used in the models will include both dosing and concentration records, relative time variables, and numeric covariate variables. {admiral} is an open-source R package for creating CDISC ADaM data. It can be used effectively to create Population PK analysis data (ADPPK). Additional tools from other Pharmaverse packages such as {metacore}, {metatools} and {xportr} can be used to simplify the workflow. I will discuss some of the challenges of Population Pharmacokinetic analysis data programming and show some of the solutions developed in {admiral} and the Pharmaverse.

Agenda

What is PK data
Types of PK analysis data
Brief overview of {admiral} and the Pharmaverse
Programming Workflow
Some Additional Pharmaverse tools
Pharmaverse Website Example

Pharmacokinetics Overview

Pharmacokinetics is the interaction of a drug with the body
Samples collected at regular time intervals after dosing
Metrics calculated from concentrations over time

Two Main Types of Modelling

Non-Compartmental Analysis (NCA)
Population PK (PopPK)

Population PK (PopPK) – `ADPPK`

Population Based
Less Intensive Sampling
Can Detect Variations based on Covariates

CDISC Standards for Population PK Analysis

Published October 6, 2023!

https://www.cdisc.org/standards/foundational/adam/basic-data-structure-adam-poppk-implementation-guide-v1-0

CDISC ADaM PK Standards

ADNCA (NCA) – released November 2021
ADPPK (PopPK) – released October 2023
ADPP (PK Parameters) – coming in 2024

Characteristics of PK Analysis Data

Includes both exposure (EX) and concentration data (PC)
Includes nominal and actual relative time variables
Duplicated Records for Analysis (ADNCA)
Numeric Covariates for Modelling (ADPPK)
May need to expand dosing records

Time Variables (`ADPPK`)

Variable	Variable Label
`NFRLT`	Nominal Rel Time from First Dose
`AFRLT`	Actual Rel Time from First Dose
`NPRLT`	Nominal Rel Time from Previous Dose
`APRLT`	Actual Rel Time from Previous Dose

Time Variables Example

`USUBJID`	`EVID`	`NFRLT`	`AFRLT`	`NPRLT`	`ARRLT`
001	0	0	-0.05	0	-0.05
001	1	0	0	0	0
001	0	0.5	0.465	0.5	0.465
001	0	3	2.89	3	2.89
001	1	24	24.05	0	0
001	0	25	25.15	1	1.15

`ADPPK` Expected Variables

EVID Event ID
DV Dependent Variable
MDV Missing Dependent Variable
BLQ Below Limit of Quantitation

`ADPPK` Covariates

<COV>BL for baseline covariate, (e.g., WTBL, BMIBL)
<COV>N for numerical version of categorical covariate (e.g., SEXN, RACEN)
<COV>I for any covariates with imputed values (e.g., WTI, BMII)
<COV>GRy for grouping covariates (e.g. AGEGR1)

`{admiral}`

ADaM in R Asset Library
{admiral} is Open Source and Collaborative
{admiral} is Modular
{admiral} is Part of Pharmaverse

https://github.com/pharmaverse/admiral

`{admiral}` Functions Used

derive_vars_dtm()
derive_vars_dtm_to_dt()
derive_vars_dtm_to_tm()
derive_vars_dy()
derive_vars_duration()
create_single_dose_dataset()

`{admiral}` Functions Used (continued)

derive_vars_merged()
derive_vars_joined()
derive_vars_transposed()
compute_bmi()
compute_bsa()
compute_egfr()

Other Pharmaverse Packages in `ADPPK` Workflow

{pharmaversesdtm} CDISC pilot SDTM data
{metacore} Store metadata
{metatools} Work with metadata and perform checks
{xportr} Perform checks and export transport files (XPT)

End to End Example Site

https://pharmaverse.github.io/e2e_pk/

Will focus on ADPPK example

Coding Highlights

Load Specs with {metacore}
Derive PC Dates
Expand Dosing Records
Find First Dose
Find Previous Dose
Find Previous Nominal Dose
Derive Covariates Using {metacore}
{metacore} Checks
{xportr} Steps

Load Specs with `{metacore}`

# ---- Load Specs for Metacore ----
metacore <- spec_to_metacore("pk_spec.xlsx") %>%
  select_dataset("ADPPK")

Derive PC Dates

pc_dates <- pc %>%
  # Join ADSL with PC (need TRTSDT for ADY derivation)
  derive_vars_merged(
    dataset_add = adsl,
    new_vars = adsl_vars,
    by_vars = exprs(STUDYID, USUBJID)
  ) %>%
  # Derive analysis date/time
  # Impute missing time to 00:00:00
  derive_vars_dtm(
    new_vars_prefix = "A",
    dtc = PCDTC,
    time_imputation = "00:00:00"
  ) %>%
  # Derive dates and times from date/times
  derive_vars_dtm_to_dt(exprs(ADTM)) %>%
  derive_vars_dtm_to_tm(exprs(ADTM)) %>%
  # Derive event ID and nominal relative time from first dose (NFRLT)
  mutate(
    EVID = 0,
    DRUG = PCTEST,
    NFRLT = if_else(PCTPTNUM < 0, 0, PCTPTNUM), .after = USUBJID
  )

Expand Dosing Records

ex_exp <- ex_dates %>%
  create_single_dose_dataset(
    dose_freq = EXDOSFRQ,
    start_date = ASTDT,
    start_datetime = ASTDTM,
    end_date = AENDT,
    end_datetime = AENDTM,
    nominal_time = NFRLT,
    lookup_table = dose_freq_lookup,
    lookup_column = CDISC_VALUE,
    keep_source_vars = exprs(
      STUDYID, USUBJID, EVID, EXDOSFRQ, EXDOSFRM,
      NFRLT, EXDOSE, EXDOSU, EXTRT, ASTDT, ASTDTM, AENDT, AENDTM,
      VISIT, VISITNUM, VISITDY,
      TRT01A, TRT01P, DOMAIN, EXSEQ, !!!adsl_vars
    )
  ) %>%
  # Derive AVISIT based on nominal relative time
  # Derive AVISITN to nominal time in whole days using integer division
  # Define AVISIT based on nominal day
  mutate(
    AVISITN = NFRLT %/% 24 + 1,
    AVISIT = paste("Day", AVISITN),
    ADTM = ASTDTM,
    DRUG = EXTRT
  ) %>%
  # Derive dates and times from datetimes
  derive_vars_dtm_to_dt(exprs(ADTM)) %>%
  derive_vars_dtm_to_tm(exprs(ADTM)) %>%
  derive_vars_dtm_to_tm(exprs(ASTDTM)) %>%
  derive_vars_dtm_to_tm(exprs(AENDTM))

Find First Dose

adppk_first_dose <- pc_dates %>%
  derive_vars_merged(
    dataset_add = ex_exp,
    filter_add = (!is.na(ADTM)),
    new_vars = exprs(FANLDTM = ADTM, EXDOSE_first = EXDOSE),
    order = exprs(ADTM, EXSEQ),
    mode = "first",
    by_vars = exprs(STUDYID, USUBJID, DRUG)
  ) %>%
  filter(!is.na(FANLDTM)) %>%
  # Derive AVISIT based on nominal relative time
  # Derive AVISITN to nominal time in whole days using integer division
  # Define AVISIT based on nominal day
  mutate(
    AVISITN = NFRLT %/% 24 + 1,
    AVISIT = paste("Day", AVISITN),
  )

Find Previous Dose

adppk_prev <- adppk_first_dose %>%
  derive_vars_joined(
    dataset_add = ex_exp,
    by_vars = exprs(USUBJID),
    order = exprs(ADTM),
    new_vars = exprs(
      ADTM_prev = ADTM, EXDOSE_prev = EXDOSE, AVISIT_prev = AVISIT,
      AENDTM_prev = AENDTM
    ),
    join_vars = exprs(ADTM),
    join_type = "all",
    filter_add = NULL,
    filter_join = ADTM > ADTM.join,
    mode = "last",
    check_type = "none"
  )

Find Previous Nominal Dose

adppk_nom_prev <- adppk_prev %>%
  derive_vars_joined(
    dataset_add = ex_exp,
    by_vars = exprs(USUBJID),
    order = exprs(NFRLT),
    new_vars = exprs(NFRLT_prev = NFRLT),
    join_vars = exprs(NFRLT),
    join_type = "all",
    filter_add = NULL,
    filter_join = NFRLT > NFRLT.join,
    mode = "last",
    check_type = "none"
  )

Derive Covariates Using Metacore

Use metatools::create_var_from_codelist()

#---- Derive Covariates ----
covar <- adsl %>%
  create_var_from_codelist(metacore, input_var = STUDYID, out_var = STUDYIDN) %>%
  create_var_from_codelist(metacore, input_var = SEX, out_var = SEXN) %>%
  create_var_from_codelist(metacore, input_var = RACE, out_var = RACEN) %>%
  create_var_from_codelist(metacore, input_var = ETHNIC, out_var = AETHNIC) %>%
  create_var_from_codelist(metacore, input_var = AETHNIC, out_var = AETHNICN) %>%
  create_var_from_codelist(metacore, input_var = ARMCD, out_var = COHORT) %>%
  create_var_from_codelist(metacore, input_var = ARMCD, out_var = COHORTC) %>%
  create_var_from_codelist(metacore, input_var = COUNTRY, out_var = COUNTRYN) %>%
  create_var_from_codelist(metacore, input_var = COUNTRY, out_var = COUNTRYL) %>%
  mutate(
    STUDYIDN = as.numeric(word(USUBJID, 1, sep = fixed("-"))),
    SITEIDN = as.numeric(word(USUBJID, 2, sep = fixed("-"))),
    USUBJIDN = as.numeric(word(USUBJID, 3, sep = fixed("-"))),
    SUBJIDN = as.numeric(SUBJID),
    ROUTE = unique(ex$EXROUTE),
    FORM = unique(ex$EXDOSFRM),
    REGION1 = COUNTRY,
    REGION1N = COUNTRYN,
    SUBJTYPC = "Volunteer",
  ) %>%
  create_var_from_codelist(metacore, input_var = FORM, out_var = FORMN) %>%
  create_var_from_codelist(metacore, input_var = ROUTE, out_var = ROUTEN) %>%
  create_var_from_codelist(metacore, input_var = SUBJTYPC, out_var = SUBJTYP)

Calculate Additional Covariates from Vital Signs and Labs

covar_vslb <- covar %>%
  derive_vars_merged(
    dataset_add = vs,
    filter_add = VSTESTCD == "HEIGHT",
    by_vars = exprs(STUDYID, USUBJID),
    new_vars = exprs(HTBL = VSSTRESN)
  ) %>%
  derive_vars_merged(
    dataset_add = vs,
    filter_add = VSTESTCD == "WEIGHT" & VSBLFL == "Y",
    by_vars = exprs(STUDYID, USUBJID),
    new_vars = exprs(WTBL = VSSTRESN)
  ) %>%
  derive_vars_transposed(
    dataset_merge = labsbl,
    by_vars = exprs(STUDYID, USUBJID),
    key_var = LBTESTCDB,
    value_var = LBSTRESN
  ) %>%
  mutate(
    BMIBL = compute_bmi(height = HTBL, weight = WTBL),
    BSABL = compute_bsa(
      height = HTBL,
      weight = HTBL,
      method = "Mosteller"
    ),
    CRCLBL = compute_egfr(
      creat = CREATBL, creatu = "SI", age = AGE, weight = WTBL, sex = SEX,
      method = "CRCL"
    ),
    EGFRBL = compute_egfr(
      creat = CREATBL, creatu = "SI", age = AGE, weight = WTBL, sex = SEX,
      method = "CKD-EPI"
    )
  ) %>%
  rename(TBILBL = BILIBL)

`{metacore}` Checks

adppk <- adppk_prefinal %>%
  drop_unspec_vars(metacore) %>% # Drop unspecified variables from specs
  check_variables(metacore) %>% # Check all variables specified are present and no more
  check_ct_data(metacore) %>% # Checks all variables with CT only contain values within the CT
  order_cols(metacore) %>% # Orders the columns according to the spec
  sort_by_key(metacore) # Sorts the rows by the sort keys

`{xportr}` Steps

adppk_xpt <- adppk %>%
  xportr_type(metacore) %>% # Coerce variable type to match spec
  xportr_length(metacore) %>% # Assigns SAS length from a variable level metadata
  xportr_label(metacore) %>% # Assigns variable label from metacore specifications
  xportr_format(metacore) %>% # Assigns variable format from metacore specifications
  xportr_df_label(metacore) %>% # Assigns dataset label from metacore specifications
  xportr_write(file.path(dir, "adppk.xpt")) # Write xpt v5 transport file

`ADPPK_XPT`

       USUBJID EVID NFRLT NPRLT       AFRLT       APRLT         DV
1  01-701-1028    0  0.00  0.00 -0.50000000 -0.50000000 0.00000000
2  01-701-1028    1  0.00  0.00  0.00000000  0.00000000         NA
3  01-701-1028    0  0.08  0.08  0.08333333  0.08333333 0.10156622
4  01-701-1028    0  0.50  0.50  0.50000000  0.50000000 0.54690177
5  01-701-1028    0  1.00  1.00  1.00000000  1.00000000 0.92546539
6  01-701-1028    0  1.50  1.50  1.50000000  1.50000000 1.18750588
7  01-701-1028    0  2.00  2.00  2.00000000  2.00000000 1.36888945
8  01-701-1028    0  4.00  4.00  4.00000000  4.00000000 1.68314758
9  01-701-1028    0  6.00  6.00  6.00000000  6.00000000 1.75529232
10 01-701-1028    0  8.00  8.00  8.00000000  8.00000000 1.77185470
11 01-701-1028    0 12.00 12.00 12.00000000 12.00000000 0.49503585
12 01-701-1028    0 16.00 16.00 16.00000000 16.00000000 0.13792316
13 01-701-1028    0 24.00 24.00 24.00000000 24.00000000 0.01070627
14 01-701-1028    1 24.00  0.00 24.00000000  0.00000000         NA
15 01-701-1028    0 36.00 12.00 36.00000000 12.00000000         NA
16 01-701-1028    0 48.00 24.00 48.00000000 24.00000000         NA
17 01-701-1028    1 48.00  0.00 48.00000000  0.00000000         NA
18 01-701-1033    0  0.00  0.00 -0.50000000 -0.50000000 0.00000000
19 01-701-1033    1  0.00  0.00  0.00000000  0.00000000         NA
20 01-701-1033    0  0.08  0.08  0.08333333  0.08333333 0.10052550

Summary

CDISC ADaM Implementation guides for ADNCA and ADPPK available
{admiral} template programs exist for these
Pharmaverse tools can streamline the workflow
Examples Website https://pharmaverse.github.io/e2e_pk/

Thank You!

{admiral} Core Team
Slack Pharmaverse PK Working Group
Luke Reinbolt
Ben Straub